with -OO.  Tests requiring docstrings are skipped.  Patch by
Brian Curtin, thanks to Matias Torchinsky for helping review and
improve the patch.

Noted by Joseph Armbruster; patch by Jessica McKellar.

The original code was 'for (;;) {...}', where ... ended
with a 'return -2' statement and did not contain a 'break' or 'continue'
statement.  Therefore, the body of the loop is always executed once.

Once upon a time there was a 'continue' in the loop, but it was removed in
rev36346, committed by mwh on Wed Jul 7 17:44:12 2004.

(I asked the BDFL first, and he approved removing it.  The last actual bugfix
to Tools/modulator was in 2001; since then all changes have been search-and-replace:
string methods, whitespace fixes, etc.)

it would call _unlock_file() and pass it a closed file object.  Noted by Rob Austein.

Issue #7988: Fix default alignment to be right aligned for complex.__format__. Now it matches other numeric types.

Clarify description of three-argument pow for Decimal types:  the exponent of the result is always 0.

Text by Alexey Shamrin; I changed 'DEC Alpha' to the more relevant 'Intel Itanium'.

server to resource in the 405 response msg.

#7076: mention SystemRandom class near start of the module docs; reword change description for clarity.  Noted by Shawn Ligocki.

complex.__coerce__.  Thanks Meador Inge for the patch.

message/rfc822 it turns it into an object whose body consists of
a list containing a single Message object.  HeaderParser, on the
other hand, just copies the body as a string.  Generator.flatten
has a special handler for the message mime type that expected the
body to be the one item list.  This fails if the message was parsed
by HeaderParser.  So we now check to see if the body is a string
first, and if so just we just emit it.

- add corresponding documentation

Fix unittest.TestCase.assertDictContainsSubset so it can't die with unicode issues when constructing failure messages. Issue 7956